Deployment of high speed transparent and reconfigurable optical networks requires effective, flexible and robust Optical Performance Monitoring (OPM) techniques in order to ensure high quality of service as well as high level of resiliency.
The adoption of optical coherent detection, in which the carrier phase and amplitude are recovered at the receiver-side and down-converted to the electrical domain (as opposed to direct detection, in which the phase information is lost), provides an additional degree of freedom to encode and transmit information and therefore a gain in spectral efficiency. Most importantly, this lossless optical-to-electrical signal conversion offers dramatic boost to the applicability of Digital Signal Processing (DSP), following high speed analog to digital conversion.
At the transmitter-side, the DSP may be used mainly as follows:                1. To implement advanced Dual Polarization (DP) modulation formats (for example DP-BPSK, DP-QPSK, DP-8QAM, DP-16QAM) in order to carry more bits per symbol;        2. To enhance spectral efficiency of multi-channel transmission systems by employing techniques such as Nyquist pulse shaping or Orthogonal Frequency Division Multiplexing (OFDM);        3. To implement pre-distortion techniques in order to enhance the resilience of signal propagation to fiber impairments; and        4. To apply software-defined modulation in order to adapt the signal to time/spatial-varying properties of the communication channel and to varying transmission capacity requirements.        
At the receiver-side, the DSP may be used mainly to:                1. Ease the requirements of the optical receiver, making coherent reception more cost effective, as complexity may be shifted from the optical domain to the electrical domain (e.g., by applying digital compensation of the frequency carrier offset and the optical phase noise);        2. Compensate distortions caused by signal propagation via the optical network (Chromatic Dispersion (“CD”), Polarization Mode Dispersion (“PMD”), Polarization Dependent Loss (“PDL”)), thus enabling to improve the transmission capacity and the reach distance;        3. Provide performance monitoring parameters such as In-Band Optical Signal to Noise Ratio (“OSNR”), Electrical Signal to Noise Ratio (“ESNR”), accumulated CD, PMD and PDL of the detected signal;        4. Adaptively reconfigure signal detection strategies in order to cope with dynamic networks;        5. Enable the use of Soft Decision Forward Error Correction (SD-FEC) techniques to enable increasing impairment resiliency.        
With the shift towards advanced coherent modulation formats and the use of DSPs, high spectral efficiency optical networks may be designed with almost no restriction on accumulated CD and PMD. Current technologies enable compensation of up to +/−60 000 ps/nm accumulated CD and 30 ps of PMD. Consequently, the transmission reach is limited mainly by the Amplified Spontaneous Emission (“ASE”) noise from the optical amplifiers and the optical nonlinear effects.
Real time monitoring of the OSNR is a requirement set to ensure satisfactory signal quality and to monitor potential failures at the transmission link. Several methods have been proposed in the art to derive the In-Band OSNR level by estimating the in band noise level directly, even in the presence of optical filters along the link. These methods comply with the use of polarization multiplexing and coherent optical modulation formats. Two methods for In-Band OSNR monitoring based on Stimulated Brillouin Scattering (“SBS”) effect have been described in the Applicant's patent applications published under US 20120063772 and US 20120219285 and are hereby incorporated by reference.
Other methods which rely upon the use of a DSP in a coherent receiver have also been proposed. For example, Z. Dong, A. P. T Lau and C. Lu, in “OSNR monitoring for QPSK and 16-QAM systems in presence of fiber nonlinearities for digital coherent receivers”, Optics Express, vol. 20, no. 17, pp. 19520-19534, 2012, describe a method for fiber-nonlinearity-insensitive OSNR monitoring in digital coherent receivers, which relies on incorporating and calibrating fiber nonlinearity-induced amplitude noise correlations among neighboring symbols into conventional OSNR estimation techniques from received signal distributions.
However, the monitoring of the OSNR level of the signal is still not sufficient in order to monitor the overall OSNR system margin. The overall OSNR system margin is defined as the margin in term of OSNR from the current operating OSNR level of the channel, to the OSNR level that is attained for a given pre FEC Bit Error Rate (“BER”) target. Usually, this is the pre FEC BER threshold for which the post FEC BER is 10−15. Link induced physical degradations, such as received optical power to the receiver, CD, PMD, PDL and more specifically nonlinear effects, can change significantly the OSNR level to be attained for a given BER target and therefore causes difficulties in the estimation of the overall OSNR system margin.
Monitoring the OSNR system margin is required in different phases of the optical network operation, starting from the link commissioning (where one needs to compare actual and expected system margin based on the network design and to proceed therefrom to the necessary adaptations if required), in-traffic operation (in order to monitor potential system degradations and to make the necessary link adaptations and/or signal rerouting if required), and in failure detection (in order to localize the fault location).
OSNR system margin monitoring is particularly beneficial when using software defined optical coherent transceivers, in order to optimize the transceiver adaptive parameters such as the bit rate, symbol rate modulation formats and FEC overhead, as part of the service and network requirements such as reach distance, capacity, service priority and latency.
A conventional prior art method for OSNR system margin monitoring is illustrated in FIG. 1. After being sent along the network link, a portion of the signal to be monitored is tapped off the link and its OSNR level is measured using an Optical Spectral Analyzer (“OSA”), followed by providing the pre FEC BER level by the receiver's FEC decoder module. Typically, in order to reach the pre FEC BER target, the signal OSNR level is deliberately deteriorated prior to reaching the receiver by using two Erbium Doped Fiber Amplifiers (EDFA) set in a cascade configuration with a Variable Optical Attenuator (VOA) that acts as a span loss element located between the two amplifiers. Such approach has the disadvantage of having to use complex and expensive network equipment that prevents its systematic use. Therefore, when required according to this method to measure OSNR margin at a given network node, one might need to physically bring these elements to the geographic location of the node (incurring significant operational expenses) and to find a monitor access point at the link where the signal may be tapped off while avoiding traffic disturbances during the measurement.
U.S. Pat. No. 7,561,797 describes a method and system for controlling OSNR of an optical signal at a receiver end of an optical link. The proposed method is based on degrading the signal at the TX side by implementing one of the following two options:
Option 1: a pre-compensating function is used for a digital filter (e.g. in order to pre-compensate the CD) while leaving a residual impairment at the receiver side (e.g. a residual chromatic dispersion).
Option 2: adding at the transmitter side, a digital electrical noise to the digital electrical signal to be transmitted before the Digital to Analog Converter (“DAC”) operates thereon.
However, U.S. Pat. No. 7,561,797 has the disadvantages that it requires to degrade the signal before its transmission via the link, and that information is required to be sent from the receiver, back to the transmitter, via a control channel in order to be able to control the degradation strength. In addition, deriving the in band OSNR only based on the pre FEC BER measurement, is not accurate enough since the pre FEC BER is proportional to the electrical Signal to Noise Ratio (ESNR) and the ESNR and OSNR are linearly proportional only when the OSNR level is low enough and when the only predominant impairment in the link is derived from the ASE noise. In case of a nonlinear impairment and/or CD and PMD impairments, the linear relationship between the ENSR and OSNR is not valid.
Therefore an accurate OSNR system margin monitoring method is required. One that should be robust to link impairments such as fiber nonlinearities, CD, PMD and PDL. Such a method should not affect the signal service quality and should enable remote monitoring operation in order for it to be cost effective.